Add http failure reason metric #1077

[domcyrus]

Adds probe_http_failures_total metric implementing feature for #1077 . This couter type metric provides the reason of http probe failures.

For example:

probe_http_failures_total{reason="dns_not_found"}
probe_http_failures_total{reason="request_creation_error"}
probe_http_failures_total{reason="request_get_timeout"}
...

[slrtbtfs]

Hi, first of all thanks for this PR!

May i ask why this PR introduces a global probe_http_failures_total metric instead of the per-probe probe_http_failure_reason metric proposed by #1077 ?

The latter seems like a better fit for the underlying issue to me.

(To avoid confusion: I'm not a maintainer, just someone who would like to use this feature.)

[domcyrus]

Hi, first of all thanks for this PR!

May i ask why this PR introduces a global probe_http_failures_total metric instead of the per-probe probe_http_failure_reason metric proposed by #1077 ?

The latter seems like a better fit for the underlying issue to me.

(To avoid confusion: I'm not a maintainer, just someone who would like to use this feature.)

@slrtbtfs Yes, looking at this again I also do think that we don't need to introduce a new global metric per-probe. Regarding the metric name in #1077 @SuperQ proposed to use probe_failure_info.

[slrtbtfs]

Ok, if you don’t mind, I’ll try to come up with a new PR using probe_failure_info instead then.